an empirical analysis of cooperative creation, survival
TRANSCRIPT
An empirical analysis of cooperative creation, survival compared to
capitalist firms and survival between different co-op types in the United
Kingdom
201174165
Supervisor: Dr Eike Mark Rinke
Submitted in partial fulfilment of the requirements for the degree of Bachelor of Arts in
International Relations with Quantitative Research Methods
POLIS
May 2020
Word count: 9,992
2
Abstract
Cooperatives are democratically controlled enterprises functioning in market economies
and are prevalent in the United Kingdom. What sets them apart from regular capitalist
firms is the fact that they relegate decision making to the entirety of their workers/owners
as opposed to only listening and considering the opinions of owners/shareholders. This
paper seeks to test three things. Firstly, it asks the question: why are cooperatives
created? Secondly, an analysis of hazard ratios and survival are presented when
discussing co-ops compared to capitalist firms. Finally, an exploratory analysis examines
the differences in survival rates and hazard ratios of different types of cooperatives. The
analysis is relegated to the United Kingdom and is empirical. Statistical methods such as
linear regression and Cox proportional hazards models are used in the empirical analyses.
The main conclusions to be drawn from the paper are threefold. Firstly, Cooperatives are
created during times of disillusionment with capitalism and unemployment. Secondly,
democratically controlled firms have lower hazard ratios and survive better in market
economies than regular capitalist firms during a five-year period. Finally, it is shown that
consumer cooperatives have smaller hazard ratios and tend to survive at higher rates than
worker co-operatives during a five-year period.
3
Table of contents
I. Introduction II. Definitions
III. Reasons for cooperative creation Literature on disaffection with the capitalist economic system and unemployment Methodology and data Empirical model Discussion: Widening the understanding of co-op creation Limitations Further research
IV. DF and CF survival comparisons
Literature on cooperative versus traditional firm survival Methodology and data Empirical model Discussion: Why do cooperatives survive at higher rates? Limitations Further research
V. Cooperatives in depth
Lack of literature on survival of different cooperative types Methodology and data Empirical model Discussion: Are industrial sectors responsible for different co-op survival rates? Limitations Further research
VI. Conclusions
VII. Bibliography
VIII. Appendix
IX. Notes
4
I. Introduction
The year 2012 has been declared the international year of cooperatives by the UN,
highlighting the positive effects they have done on communities and economies (Cheney
et al. 2014). What is interesting about these organisations is that they differ from what we
traditionally in our capitalist society would consider to be a company. Rather than the top-
down solution to business, where bosses and shareholders control the enterprise, co-ops
emphasise democratic control and a bottom-up approach to managing a business. The
purpose of this study is threefold: firstly, it aims to explain why a person would embark on
a venture of creating a co-operative with variables such as disaffection with the capitalist
system as well as unemployment playing a positive effect on creation of co-ops. Secondly,
the survival and hazard rates of cooperatives (DF) and traditional capitalist firms (CF) is
analysed to determine which type of company survives better in the market economy.
Finally, an exploratory analysis on different co-op types and their survival and hazard rates
is analysed to determine the most successful types of co-operatives and why they survive
longer than their counterparts. The analysis for all three sections is empirical and relegated
to the United Kingdom. This paper also seeks to contribute to the lack of literature in the
field of research on capitalism and cooperatives by offering confirmatory analyses in the
first two sections and exploratory analysis in the third section. An empirical section will
offer a literature review first, followed by the methodology and data descriptions, empirical
results, discussion, limitations and suggestions for further research. All of the data
manipulations and analyses have been done with Stata 16. To allow for scrutiny and
replication of the analysis, all of the datasets and code are available online1 (Anonymous,
2020).
1 For the direct link to the depository see Notes
5
Definitions
To begin with an accurate analysis of cooperatives, one has to define them. For this paper,
the definition of a co-op is based on the definition given by the International Cooperative
alliance (n.d.):
An autonomous association of persons united voluntarily to meet their
common economic, social and cultural needs and aspirations through a
jointly-owned and democratically controlled enterprise.
This shall be the working definition of cooperatives as an umbrella term for democratically
controlled enterprises. A traditional business then is defined by a lack of democratic
control of the members/workers of the enterprise. To further this definition, a traditional
business is also considered to be owned by its shareholders/leaders, who are private
individuals (Barkley Rosser Jr and Rosser, 2004, p.7). This is done in order to clear up any
misreading of the study as some may claim that because co-ops as well as CF’s function
in market economies dominated by private property, that would make them capitalist as
well. This is because of the widespread mis(conception) that free markets are only
possible in capitalist economies, however this way of thinking is inherently flawed, as free
market economies have been advocated by anti-capitalist theorists as well (Bockman,
2011). The study seeks to highlight the differences in organisational structure, either
democratic or capitalist. Of course, it must also be mentioned that amid worker-owners in
co-ops there may be a minority of non-owning hired workers (Ben-Ner, 1988, p. 8). Even
so, since only a minority of hired workers do not have decision power over the firm it does
not diminish the distinction between a firm where the majority have sway over what
happens to it and one where only a small fraction of shareholders and owners dictate the
direction of the company.
6
II. Reasons for cooperative creation
Cooperatives are an organisational alternative to traditional capitalist firms, however, what
motivates someone to embark on a venture to create a co-op as opposed to a CF? There
are, of course a myriad of them and the full discussion of the question is beyond the scope
of this paper. Rather, the reasons discussed in this section will be more ideological and
economical in their nature: a growing dissatisfaction with the current capitalist system and
an increase in unemployment.
Literature on disaffection with the capitalist economic system and unemployment
The theoretical conception of why a cooperative would be created can be derived from the
seminal works of Karl Marx (1848;1867). In the Communist Manifesto Marx (1848, p.14)
proposes the idea that the history of all of society is the history of class struggle. It is
assumed that by and large the capitalist economic system is divided into two opposing
classes: the bourgeoisie (owners of the means of production and employers of workers)
and proletarians (those who work for the capitalists and do not own the means of
production). This relationship is categorised as exploitative by Marx (1848;1867) as
capitalists extract surplus value from the workers, depriving them of the full value of their
labour to expand their own gains. The exploitation leads to antagonisms, which pit the
worker against the capitalist. The way, then, for the relationship to be absolved of
antagonisms is for workers to own the means of production themselves. This is where the
link between Marxist theory and cooperatives intersects, as the voluntary association and
democratic member self-management of co-ops gives them more and sometimes even full
control over the workplace, meaning that the workers are in control of the means of
production. The antagonisms that lead to the unrest of the workers, however, are
controlled by other ways in capitalist economies To begin with, there are material aspects
7
that states can utilise such as creating social safety nets to the majority of the working
population2 or outright military crackdown on persons who seek to have a democratic say
in the economy (Lenin, 1917). There are also ideological aspects of control that capitalist
societies use to entrench their ideologies. The one discussed in this paper has been
coined by the late theorist Mark Fisher (2009) in his seminal work Capitalist Realism: Is
There no Alternative?. The term “Capitalist Realism” as defined by Fisher (2009, p.8) is
The widespread sense that not only is capitalism the only viable political
and economic system, but also that it is now impossible even to imagine a
coherent alternative to it.
Examples of this way of thinking can be seen in the work of Fukuyama (1989), who
theorises that after the fall of the Soviet Union, the ideological evolution of human societies
has ended, with democratic Western liberal capitalism coming out on top as the universal
and best human system. Indeed, this way of conceptualising the world is diagnosed as the
realism that pervades out societies today by Fisher. However, it does not have to be the
only way to conceptualise politics and the economy. Fisher (2009, p. 22) writes that one
way that capitalist realism can be threatened is to show that it is inconsistent or untenable.
One of the ways it can be made vulnerable is via crises, the main example here being the
2008 financial crisis, which engulfed most of the world and was compared to the 1929-
1933 Great Depression in terms of its magnitude and effects (Eigner and Umlauft, 2015).
The aftermath of the financial crash of 2008 in the UK resulted in the newly appointed
Conservative-Liberal Democrat coalition imposing austerity measures which, the working
persons believed, disproportionately affected their livelihoods in the negative sense
(Mustafa, 2011). The dissolution of social safety nets that served to sustain working
2 These allow for the majority to live comfortable lives and as such mitigate the antagonisms of exploitative capitalist working conditions.
8
persons enough so they would not feel that their relationship towards the rulers of capital
was inherently antagonistic had begun to be eroded away and as such, alternatives could
begin to flourish in the economic sphere and capitalist realism challenged with that. One of
these alternatives could of course be forming a cooperative that provides democratic
control over the workplace for its workers, mitigating class antagonisms and giving a voice
to those who felt disenfranchised by lack of control over their workspaces. Given the
analysis presented by Fisher and Marx and the crisis that had recently affected the United
Kingdom I hypothesise that more people will break away from capitalist realism, forming
cooperatives as an alternative mode of economic organisation during the years that the
crisis is taking place, and after it. This is due to the system that contains class
antagonisms failing to do so as well as the imposition of austerity measures after the crisis,
further eroding capitalist realism as well as creating a disaffection with dominant mode of
production, which leads to an increase of cooperative creation as an alternative. The first
hypothesis then is:
H1: During crisis years and the years after a crisis the number of created cooperatives will
increase
The increase in unemployment that accompanied the crisis can also be another factor in
facilitating cooperative creation. As more and more potential employers are put out of
business, the more workers are left without jobs and some could decide to band together
in the formation of a DF whilst sharing the initial costs of setting up a business together by
either starting the firm from scratch or buying it from their former employer after the firm
has bankrupted. Indeed, this is supported by empirical literature. Ben-Ner (1988, p.9-10)
shows that the relationship between unemployment increases and the increase in the
number of worker-owned firms created is positive. This is further examined in the study by
also showing how many new conventional capitalist firms were created as unemployment
9
was going up. The relationship is again positive, however, comparatively the growth for
cooperative numbers is much higher than for CF growth (Ben-Ner, 1988). This idea,
though, has not been touched upon much in empirical literature. For this, it is imperative to
check the validity of such claims and to try to replicate the limited finding of empirical
literature. The formulated hypothesis is:
H2: Increases in unemployment have a positive relationship with the increase in number of
co-ops created
Methodology and data
To test these hypotheses a linear regression model has been created to observe how
crisis years as well as unemployment increases impact cooperative creation during the
timeframe between 2000 and 2018 with both years included. The linear regression model
has been chosen because in the Ben-Ner study (1988, p 9-10) the increases in
unemployment and co-op creation seem to have a lineal positive relationship. For
disaffection this is also acceptable, since there has not been any empirical literature
surrounding the question as of the writing of this paper, and as such there is no
justification for supporting complex models to analyse the first hypothesis. As such, the
simplest OLS model was judged adequate for the study. The data for unemployment UK
has been gathered from Statista (Clark, 2019). The data for co-ops has been collected
from their open data portal (Co-operatives UK, 2019a). Table 13 shows the variables in
depth. The years during crisis have been chosen as 2008 and 2009 due to the fact that in
the UK the recession was active during both of them (Office for National Statistics, 2018).
It is argued that the disillusionment with capitalism is at its baseline value (0) during the
years before the crisis, with it increasing more during the crisis itself (1) and culminating in
3 See Appendix
10
the years after the crisis (2) as more and more people get affected by the economic
hardships caused by the current economic system’s failures to provide adequate working
conditions and living standards. As both explanatory variables could suffer from the
multicollinearity problem with unemployment being a cause for disaffection with capitalism
as well as years after crisis predicting unemployment increases, diagnostic measures such
as the Variance Inflation Factor (VIF) have been implemented for the model to test for the
increase in variance that possible multicollinearity may create. To test for the accuracy of
the model, prediction errors and out of sample errors for the model are presented later in
the empirical results.
Empirical model
Table 2. Unemployment-Disaffection model for cooperative creation
B SE VIF p-value
Unemployment increase 32.25 6.47 1.22 0.000
Disaffection with
capitalism
52.12 8.76 1.22 0.000
Observations 19
R2 0.867
RMSE 32.63
Before discussing the model in depth, it should be noted for an easier interpretation that
the mean number of cooperatives created during a year is 224 when rounded, with the
biggest number of co-ops created being 374 during the year 2013 and the smallest
number being 110 in 2006. Table 2 shows us that the model gives support for both
hypotheses with both independent variables showing a positive relationship between them
and an increase in the number of created cooperatives. Both values are highly statistically
significant (p<0.001). A one percent increase in unemployment is associated with a
creation of 32 extra cooperatives during that year. In regard to the disaffection with the
current economic system it is found that a one unit increase in dissatisfaction with
11
capitalism is associated with 52 extra cooperatives created during that year. This means,
that in the years after the crisis 104 extra cooperatives are expected than in the years
before the recession which is a 46% increase when compared to the mean of co-ops
created during an individual year. Diagnostics on the model have also been performed to
see how it performs. The Variance Inflation Factory (VIF) for both of the variables is 1.22
which shows a slight increase in the variance of the model because of independent
variable collinearity. However, this is not a huge problem as models are deemed to be
problematic if VIF for the variables exceeds 5 or 10 (Kutner et al. 2004; Sheather, 2009).
The given R-squared value suggests that the independent variables explain around 87%
of the variance of the dependent variable, showing a good fit of the model. The root mean
squared error (RMSE) has also been presented to show the amount prediction error
present in the model. To analyse this more closely one must look at Table 3 as it shows
the prediction errors and out of sample errors for each individual observation.
Table 3. Individual Prediction and Out of sample errors for the Unemployment-Disaffection model
Year Unemployment
(%) Crisis Year(1) Prediction Error Out of Sample
Error
2000 5.4 B -18.83 -21.38
2001 5.1 B .84 .95
2002 5.2 B 39.62 44.92
2003 5 B 24.06 27.35
2004 4.8 B 31.51 36.03
2005 4.8 B 2.51 2.88
2006 5.4 B -44.83 -50.9
2007 5.3 B 1.39 1.58
2008 5.7 D -32.63 -34.47
2009 7.6 D -39.91 -48.63
2010 7.9 A -54.71 -68.06
2011 8.1 A 26.84 34.49
2012 8 A 27.07 34.2
2013 7.6 A 43.97 52.59
2014 6.2 A -3.88 -4.35
2015 5.4 A 36.92 43.11
2016 4.9 A -21.96 -27.18
2017 4.4 A 6.17 8.34
2018 4.1 A -24.16 -35.04 (1)B-Before Crisis, D-During Crisis, A-After Crisis
12
It can be seen that the model performs reasonably well at predicting the number of
cooperatives that will be created each year. It is worth noting that the model predicted
more cooperatives created during crisis years than there actually were. This can be
explained in the sense that the effects of the crisis were not fully present, with austerity
measures starting to be imposed during 2010 (Muller, 2019). This assumption has further
grounding if one looks at the years after the imposition of austerity (2011-2018) with more
cooperatives being created during those years (except 2014, 2016 and 2018) than
predicted. It may be that unemployment and disaffection with capitalism may play a bigger
role in cooperative creation than shown by the model, or it could be a myriad of other
reasons that were not considered in this model affecting this increase in co-ops.
Discussion: Widening the understanding of co-op creation
There are multitudes of reasons that co-ops can and are formed in market economies, and
disaffection with the capitalist mode of production and unemployment increases do seem
to affect the creation of co-ops. Whilst the findings suggest a positive relationship, there is
a little amount of literature that tackles with both hypotheses. For disaffection it is only a
theoretical and historical analysis that has been put forth (Marx, 1848;1867, Fisher, 2009).
In the Marxian sense, class antagonisms do seem to cause uproars during periods of
economic hardship, which the 2008-9 crisis exemplified for the United Kingdom. As such, it
is reasonable to assume that as co-ops do provide more worker control and are to some
extent opposed to the capitalist mode of production that workers would choose to join or
create a DF in order to either advance their own interests of wanting a more egalitarian
workplace where they are able to foster change via their own vote and have more say on
day to day matters. Ideological aspects can also be conceptualised as there may indeed
be those who consider themselves fundamentally opposed to capitalism and could not in
the moral sense work in a firm where it is the shareholder-owners making all of the
13
decisions. Opposition to authoritarian4 rule of corporations is not only a Marxian stance;
democratic theory also approaches this issue on the side of the cooperatives. Dahl (1985,
p.135), one of the most important democratic theorists of our time, gives his ideas on the
matter:
Members of any association for whom the assumptions of the democratic
process are valid have a right to govern themselves by means of the
democratic process (…) We therefore see no convincing reasons why we
should not exercise our right to the democratic process just as we have
already done in the government of the state.
It would make sense that both ideological camps would be interested in creating co-ops
but for different reasons: Marxists because it is a way to subvert the capitalist economic
structure and democratic theorists because it is a way to extend democratic privileges to
the economic sphere as well as the political. Both of these ideological camps can be filled
by newcomers when the capitalist realism of Fisher (2009) is broken down by crises that
shows for some that the capitalist business is an unsustainable idea and therefore one
should seek alternatives to it. Persons could become Marxist in this sense when they are
conscious of their exploitation by the capitalist class as layoffs and decreases in wages
show that their needs are not being met by the ruling elite of a firm. For the same reason
others could be made into companions for democratic theory by crises in the sense that
they show just how little influence a lower or middle class person has when a business
fails because of mismanagements of the global market economy by the minority of
owners. This could lead a person to believe that democracy is essential in the economic
sphere like in the political sphere for an inclusive society as well as better management of
4 Authoritarian in the sense that the persons at the top make all of the decisions in a capitalist firm with little to no input from workers
14
the world economy. There are also counter arguments that could be made. Perhaps co-
ops are created at higher rates after the crisis because of recognitions from world
renowned organisations? After all, the UN declared 2012 to be the International Year of
Cooperatives and gave them lots of attention, holding conferences and publishing various
research papers about DF’s (Cheney et al. 2014). This, of course, only feeds into the
argument as declaring a year to the co-ops after an economic crisis could be one of the
ways in which capitalist realism is challenged by the UN acknowledging that alternatives to
capitalist modes of production have contributed significantly to our welfare. It does seem
then that theoretically disaffection with capitalism does indeed foster a growth in
cooperative numbers, however, there do not seem to be any empirical studies as of the
writing of this paper specifically focusing on disaffection with capitalism and creation of
cooperatives. As this is the case, it is important to conduct more studies on this topic to
see if the results of the hypotheses hold up to scrutiny and are replicable. For
unemployment increases one again finds a shortage of empirical literature to draw upon,
though this study does find evidence for ideas put forth in Ben-Ner’s (1988) work that
unemployment does indeed play a factor in facilitating the growth of the numbers of co-op
enterprises. The idea makes sense intuitively as the more workers get laid off, the more
they could be able to organise together to form enterprises or buy failing ones from their
former bosses during times of economic hardship. The effect is increased by the recent
advancements in availability of education have increased the entrepreneurial skills of
regular workers (Ben-Ner, 1988, p.22). Even so, capital supply constraints for co-ops are
prevalent (Perotin, 2004, p.83). This effect would be increased during crises years when
one could expect a high unemployment as many of the unemployed workers would not
have capital saved up or a consistent way to get more due to the very nature of being
unemployed.
15
Limitations
Even though the theory is consistent with the data, there are aspects of the model that
could be improved in future research. Firstly, direct variables, gauging disaffection with
capitalism should be used. To my knowledge, there exists no survey that deals with that
question in the UK and has been done during the timeframe that this model was run. The
model could also have benefitted from control variables like the wealth of the UK,
availability of education to workers and their education level to control for co-op creation.
This would have allowed to more accurately predict the results as well as test some of the
other claims about co-op creation in the research literature.
Further research
As this is a relatively unresearched area, the first idea would be to conduct more empirical
and theoretical studies on this topic in order to see if the proof for the hypotheses can be
replicable. For variables, it is imperative that surveys about the disaffection with capitalism
conducted every year be done in order to provide more accurate disaffection variables.
Furthermore, studies that control for other reasons for co-op creation are welcome to test
the validity of the results and to come to more accurate conclusions.
III. DF and CF survival comparisons
As the reasons for co-op creation have been discussed, it is time to see how they are able
to survive versus traditionally managed firms. This section deals with the hazard ratios and
survival rates between DF’s and CF’s
Literature on cooperative versus traditional firm survival
Better cooperative survival and lower hazard rates are backed by empirical literature,
however it mostly seems to focus on worker cooperatives. Ben-Ner (1988) conducted a
16
study that compared worker cooperatives and capitalist firms in the United Kingdom during
the timeframe of 1974-1986. An estimation of hazard rates for both types of firms was
made. It was found that worker cooperatives were much more likely to survive at all age
points than traditional capitalist firms. For Perotin (2004, p. 78) the focus was on a cohort
of French firms. Hazard rates were again estimated for both types of firms with worker
cooperatives coming out on top again with a survival rate of 75% compared to 60% of all
French firms combined after the initial four years of creation (in the fifth year following their
establishment). A more extensive study in terms of categories was done by Monteiro and
Stewart (2015) which examined cooperatives compared to capitalist firms in Portugal. The
studies show that at all age points during a five-year period, cooperatives enjoyed greater
survival chances than their capitalist counterparts. To be more precise, Monteiro and
Stewart (2015, p. 110) estimated that around 97% of tested cooperatives had survived 5
years or more, 84% for 20 years or more and 63% had survived for 50 years or more. For
traditional firms the respective percentages were approximately 80%, 45% and 20%. This
was explained by showing that consumer and supplier cooperatives had lower exit
probabilities than CFs, which contributed to their increased survival chances. However,
when going into more depth of the different categories it was found that worker
cooperatives had similar exit probabilities compared to capitalist firms though this was
based on a small sample of worker co-ops. A more explanatory empirical study was done
when comparing DFs and CFs in Uruguay by Burdin and Dean (2009) which stipulates
that the enterprises differ substantially in their economic behaviours with increases in profit
corresponding to bigger increases of wages for cooperatives than capitalist firms as well
as that cooperative employment is more resistant to price shocks than in their capitalist
counterparts. It offers empirical evidence to explain why cooperatives are more resilient,
as higher wages and more stable employment opportunities can be seen as incentives for
17
why a person would choose to continue to work in a co-op rather than go to a capitalist
firm. This could give members more enthusiasm to keep the company going. The
difference in survival rates is further contrasted by Audretsch (1991, p.443) who found that
around 50% of all businesses in the US die before the age of 6. Similar figures were
reported in the UK by Disney et al. (2003). After review of the academic literature it can be
seen that it suggests that cooperatives are better at surviving after their birth than
traditional capitalist enterprises. It is noted, that most of the papers focus on worker
cooperatives with some adding consumer firms to the analysis and as such are limited
when looking at the entirety of DF and CF comparisons. The hypotheses are:
H1: Cooperatives will have lower hazard rates than traditional capitalist firms in the first
five years of their existence
H2: Cooperative survival rates for five years will be between 75 and 97 percent
H3: Traditional firm survival for five years will be between 50 and 80 percent
The first hypothesis is based on the conclusions of (Ben-Ner, 1988; Perotin, 2004), both of
whom conclude the survival rates for cooperatives are higher than those for traditional
capitalist firms. The second hypothesis is based on Prerotin (2004) estimating co-op
survival at around 75% after five years of existence and Monteiro and Stewart (2015)
whom conclude that the DF survival rate after five years is 97%. As such, I believe that
cooperative survival in the UK will be between those two numbers. The third hypothesis is
based on Audretsch (1991) and Monteiro and Stewart (2015) who placed traditional firm
survival rates after 5 years of existence at 50 and 80 percent, respectively. Thus, I believe
that in the UK the survival rates will be between 50 and 80 percent.
Methodology and data
18
This section examines the survival rates of cooperatives and traditional capitalist firms for
enterprises born in the year 2012 through the time frame of the years 2012-17. The Cox
proportional hazards model has been chosen as the go-to model for this analysis. It is a
regression model that tests the association between survival time of an individual (in this
case a business) and one or more predictor variables (Cox, 1972). The model produces a
hazard ratio that is to be interpreted as a weighted average of the true hazard ratios over a
selected time period (Stensrud and Hernan, 2020). This is more complicated than regular
OLS or logit regressions and has been chosen because I specifically want to test the
hazards involved in different business types, which the simpler regression models simply
cannot do to the accuracy that the Cox model can. Discussing the data, it must be
mentioned that enterprise births and deaths have been defined in accordance with the
British ONS definition that was used for the 2017 business demography report (Office for
National Statistics, 2019). A birth is identified as a business that existed in the year t but
was not present in the year t-1 or t-2. A death is identified as a business which was active
in year t but was not active in year t+1 or t+2. As such, a business is deemed to have
survived the year if it was active and/or born in the year t and still active in the year t+1.
Data for cooperatives have been gathered from the Co-Operatives UK open organisational
datasets (Co-operatives UK, 2019a) and data for traditional firms has been gathered from
the ONS business demography data (Office for National Statistics, 2019). Some
corrections have been made to the merged dataset as some cooperatives can are labelled
as businesses and as such appear in the ONS data. These comprised 90 cooperatives
and they have been removed from the traditional company data5. After cleaning the
5 Survival times for these cooperatives was checked and 90 of the individual businesses which had the same survival times were removed from the ONS data
19
dataset, the number of DF’s in the dataset is 370 and the number of CF’s is 269,160.
Variable descriptions and coding can be seen in depth in Table 1.
Empirical model
An interesting note before beginning is that when applying the definitions of firm survival
rates to the cooperative data it was found, as demonstrated by Table 4, that the entire
sample of democratic firms survived their first year. This is quite interesting as the report of
Co-operatives UK (2019b, p.4) stipulates that 95.6% of the sample had survived their first
year of existence.
Table 4. Survival by firm
type
Survival Time (years)
CF DF
0 23,338 0 1 46,759 16 2 39,063 18 3 24,230 16 4 19,560 34 5 116,210 286
Total 269,160 370
Table 5. Cox porportional hazrds model on firm survival
Hazard ratio
Cooperative 0.312***
(0.034)
Observations 269530 Exponentiated coefficients; Standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001
This could be a mistake on the part of Co-operatives UK in the sense that they may have
calculated the survival rates differently than in the ONS data. As the nuances of the data
have been discussed, it is time to move on to the model itself. From Table 5 it can be seen
that the hazard ratio is much lower on average for cooperatives compared to capitalist
firms (a 69% difference). The result is highly statistically significant (p<0.001). It seems
20
then that the first hypothesis is proven by the Cox proportional hazards model:
Cooperatives do indeed have lower hazard rates than traditional capitalist firms. A point of
note here is that testing the proportional hazards assumption yielded mixed results. The
proportional hazards test was statistically significant, which suggests a violation of the it,
however when illustrating it graphically, another trend emerges. When comparing the
Kaplan-Meyer observed survival curves to the Cox predicted survival curves it can be seen
that the predicted survival curves are close to the actual survival curves, which would
mean that the proportional hazards assumption cannot be rejected.
Figure 1. Kaplan-Meyer curves compared to Cox predicted survival curves
21
Figure 2. Log-log plot of survival probability
Figure 3. Cumulative hazard rates for both types of firms
22
Figure 4. Smoothed hazard estimates for both types of firms
Figure 5. Kaplan-Meier survival estimates for capitalist firms and cooperatives
23
The results are demonstrated by Figure 1. The graphical inaccuracies between the
observed and predicted survival probabilities for cooperatives may be affected by the small
sample size.6 The proportional hazards assumption seems to have not been violated when
graphing survival probability on a log-log plot as well. Figure 2 shows this, and as the lines
of survival probability for both firms seem to be roughly parallel, which suggests no
violation of the proportional hazards assumption. More information on firm survival can be
found looking at Figure 3, which illustrates the cumulative hazards of the two types of
companies over a five-year period. It can be seen that the cumulative hazards of CF’s after
five years of creation is around 0.77 and for DF’s it is 0.25. These numbers show the
probability of the business failing over the five-year period for each subject one or multiple
times (Cleves, et al. 2008, p.15). Seeing as it is unlikely for one business to fail multiple
times within a five-year span, it would be better to interpret these findings as a probability
for a capitalist firm to survive its first five years being at 0.23 and 0.75 for a cooperative.
This shows that democratically controlled firms have a much higher probability at surviving
the first five years than their traditional counterparts. To go into more depth concerning
hazard rates one must look at Figure 4. In it, the smoothed hazard estimates for every
year which can be seen with the hazards being higher at all age points for conventional
companies when compared to democratically owned ones. However, different trends in
hazards for both types of companies can be seen. For capitalist firms the hazard increases
until year three and begins to decrease after that with the hazard rate for the fifth year
becoming similar to that of year one. This is in sharp contrast to Evans (1987), who
proposes that the risk for firms to fail decreases with age. The data here shows that the
hazard rates do not increase or decrease monotonically for conventional firms but instead
6 At the time of writing of this paper there seems to be no way to go back in time and increase the co-op sample size for a more accurate analysis.
24
form a sort of inverted U-shape. This has been described by scholars such as Bruderl and
Schussler (1990), who have found that the risk for a firm to fail increases during the first
few years of existence and then decreases. For cooperatives Figure 4 shows lower overall
hazard rates for cooperatives compared to traditional firms, with risk increasing from years
two to four and decreasing slightly during year five. Year one is excluded, since the entire
sample of cooperatives survived it. The hazard rates for DF’s are similar to the ones
produced by Ben-Ner (1988, p.19) except that in the hazard rates for cooperatives
decreasing after year three rather than after year four as is in the current data. It can be
seen that the current data on hazard rates are on a large part consistent with the existing
literature on company survival. The data also proves the first hypothesis, which suggests
that cooperatives have a lower hazard rate than traditional capitalist firms in the first five
years of their existence. Moving on to the second and third hypotheses. The Kaplan-Meier
survival curves for both company types are illustrated by Figure 5. From it, it can be seen
that 43% of traditional companies survived the initial five-year period since their creation,
whereas 77% of cooperatives did. This proves only the second hypothesis, with survival
rates for coops being between 75% and 97%, but it does not prove the third hypothesis,
with even the most pessimistic prediction of company survival (50%) being higher than the
actual survival rate of traditional companies by seven percent. This is not very surprising
as the hazard rates discussed previously indicated a major difference in risks for traditional
businesses when compared to cooperatives. In conclusion, it has been shown that
hypotheses one and two have been proven by the given data, with hazard rates being
considerably lower for cooperatives than traditional businesses as well as cooperative
survival after five years being 77% which is consistent with the current literature.
Hypothesis three, however, was rejected with projections of conventional company
25
survival shown in the research literature being higher than the actual survival rates for UK
capitalist firms.
Discussion: Why do cooperatives survive at higher rates?
It has been shown that co-ops have lower hazard ratios and higher survival percentages
than traditional firms, this section seeks to explain why such differences exist as a 34%
difference between survival rates is rather high in the business world. If capitalist firms had
the same survival rate as co-ops, that would have resulted in around 91,500 more
businesses surviving in the UK over a five-year period if they were started in 2012.
The first explanation is increased productivity in cooperatives. The logic goes that since in
market economies profit is one of the deciding factors in business survival, firms whose
workers7 are more productive will accumulate more money, allowing them to survive by
not going bankrupt. A meta-analysis of 74 studies on labour managed firms8 and
participatory capitalist firms9 by Doucouliagos (1995) found that worker participation, profit
sharing and worker ownership contributed were positively correlated with productivity in
democratically owned companies in the majority of cases. For participatory capitalist firms,
correlations between productivity and the independent variables were smaller, sometimes
even negative and less statistically significant than for democratically owned firms. This
showed that DF’s outshined CF’s in terms of productivity which was correlated with
increased worker participation in a firm. It must be kept in note, that firms with high labour
union influence and capitalist firms with no participatory measures for their employees
were not included in this study. However, given that labour managed firms beat CF’s with
7 Workers in this section are used to define all employees of cooperatives, not just those for the category of “worker cooperative” 8 Defined as a company in which all of the workers are able to participate in managing the firm and have one vote per person (Doucouliagos, 1995, p.59) 9 Defined as a company which are hierarchical in structure but implement one or more participation schemes for its employees (Doucouliagos, 1995, p.59)
26
participatory measures when it comes to productivity, it would not seem likely that CF’s
which do not allow workers to participate in its management would have increased
productivity. Secondly, it was shown by Doucouliagos (1995) that these correlations are
positive, but not very high, showing that the effect of worker participation one productivity
is positive, but does not tell the whole story when talking about productivity differences
between the firms. The positive correlations, however, can be explained theoretically. The
correlation between productivity and workplace democracy can be understood as having a
causal link also in the sense that the elimination of antagonisms between workers and
owners in the workplace creates a better place for labour and allows for a more productive
work environment. This is explained very well by Craig and Pencavel (1995, p. 125) who
show that in regular companies, owners can and do choose policies which enhance the
welfare of the owners/shareholders at the detriment of workers. The idea is further
analysed by Dahl (1985, p. 100, 105, 109) who argues that an owner of a company with
their high wages, pensions and benefits is unlikely to experience any negative effects on
their person from a bad economic situation on mismanagement of the business. Thus, it is
very probable that owners in capitalist firms are more likely to act irresponsibly,
endangering the workers they currently have employed, furthering the divide between
owner and employee. Employee owned companies, by their design, remove this conflict
and as such, worker cooperation and the collaboration between workers and management
becomes more positive and humane, leading to an increase in productivity. This is
demonstrated by Greenberg (1986, p. 41) who concludes that workers in a democratically
controlled firm are more likely to cooperate than those in a conventional business and to
help each other manifesting in more fluid work boundaries where one worker will help
another with a task to facilitate the collective growth of the workplace. This makes sense
as an individual would have more interest in furthering a collective in which the increase of
27
profit goes directly to the worker rather than to a shareholder not directly involved in the
production process. These factors in turn create more incentive to work and therefore
increases productivity, allowing the business to thrive longer.
Worker management may not only be linked to productivity, but also to better management
of the firm during economically turbulent years, increasing the likelihood of the firm
surviving. An analysis of DF’s and CF’s in Uruguay has shown that when faced with
economic hardships, co-ops and capitalist firms behave differently in regard to economic
shocks (Burdin and Dean, 2009, p. 526). Whilst a crisis for both types of companies results
in a decrease in wages and employees, capitalist firms suffer more by having a much
higher level of layoffs due to their inabilities to establish credible commitments between
workers and owners. Cooperatives, by contrast, do not suffer as many layoffs due to their
unique control structure that can employ more egalitarian ways of dealing with crises. One
of these is demonstrated by Zirakzabeh (1990, p.118) who shows that during economic
crises, workers vote voluntarily to reduce their own wages in order to sustain their
companies’ profitability and long-term survival. This allows for the collective to continue
working productively with minimal layoffs during crisis situations and thus, keep the
company going. The same issue can be viewed from a psychological lens through the
social identity theory. Social identity theory is defined as a framework that analyses in-
group bias through the idea that individuals seek to increase their self-esteem via
memberships in various social groups (e.g. family, social or economic class, country)
(Tajfel and Turner, 1979). This can create conflicts between various social groups, if one is
perceived to be dangerous and/or adverse to another. In this, instance, the place of
conflict is the workplace and the two social identities at play are economic classes – the
workers of a company and its owners. In capitalist firms, these two identities are mostly
non-interchangeable. Only the owners may be considered workers and owners since they
28
own the company and some can also do work for it, however the workers in the company
do no own it and as such are absent from categorising themselves as both owners and
workers. Of course, it is worth mentioning that not every person may choose to identify
themselves with one or both of these classes, though literature exists that confirms the
idea that class distinctions are present in workplace relationships (Marx, 1848;1867, Dahl,
1985). This distinction between worker and owner in a CF creates conflicts and animosity
between the two camps. Consider the reduction of wages or layoffs from the previous
discussion. In a social identity perspective, layoffs from a capitalist firm would create two
adverse effects: one in the relationship between classes and one between the worker
class itself. For the former an animosity might arise between worker and owner due to the
fact that workers have no say in who gets laid off and cannot vote for other measures of
tackling crises and/or reducing wages. Therefore, the owner class is perceived to be
dangerous by employees, who see themselves and their group as being negatively
affected. For inter-worker relations layoffs and wage reductions have a similar effect. Since
during a crisis the scarcity of resources is exacerbated, and with layoffs and reductions of
wages being independent of worker control, alienation between the workers themselves
occurs. In essence, a Hobbesian war of all against all becomes the norm between
employees as each seeks to survive and is more likely to do damage to the other in order
to get ahead due to limited workspaces and wage benefits. By contrast cooperatives
harmonise the identities of worker and owner by making each member contribute both
labour and capital into the enterprise. Thus, animosities are minimised during crisis years.
Democratic ownership means that members would be much less likely to support layoffs
as they themselves would be in danger, instead preferring other methods, that ensure
minimal loss of employment, such as the reduction of wages to all (Zirakzabeh, 1990,
p.118).
29
Limitations
It does seem that the data can be explained theoretically, though the empirical model itself
could be improved for the future. Firstly, the absence of control variables such as profit
and number of employees/members could not be added into the model because they were
not present in the ONS business demography data. Because of this, there may be
inaccuracies in hazard rate calculation of both companies. Secondly, the model showed
mixed results about the proportional hazards assumption, which is a core concept of the
Cox model. On the one hand, the mathematical test showed that the assumption is
violated, however, when assessed graphically, it was shown that there seems to be no
violation of the assumption. This can make the results of the model suspect, but more data
on this is required to see if it is truly the case and the results are consistent with the
academic literature on the subject.
Further research
As it stands, the results of the model show a very significand difference between co-op
and CF survival and hazard rates and as such, it is imperative, for both the economy and
science to conduct more studies on this topic as higher survival rates and productivity for a
company should be in every societies’ best interest as it could guarantee more stable
employment and generally better working conditions for employees. The area of co-op
research is still under-developed in economics and more research on this topic would
allow for a better understanding of DF’s and through that, reforms to our economic
structure may take place.
30
IV. Cooperatives in depth
It has been mentioned that cooperatives do not follow one uniform structure (save for the
fact that they are democratically managed and/or owned) or purpose. In this section the
study looks at the implications of co-operative survival based on various organisational
types. The organisation co-op dataset lists nine categories of ownership classification in its
dataset which are defined as follows by the Co-operatives UK (2019b) open data guide:
Co-operatives – members of cooperatives who are in themselves cooperatives (Co-
operatives UK).
Community of Interest – Individuals or organisations which share a common interest or
characteristic that defines their membership (Energy coops, community shops, supporters
trusts).
Consumers – Cooperatives in which members are individuals who purchase goods or
services from the co-operative itself (Consumer and retail societies, social clubs, sports
clubs)
Employee Trust – Members are employees with at least 75% stake in ownership via a trust
or similar legal entity (Most often but not always employee buyout schemes).
Enterprises – Members are legal entities, this excludes self-employed persons, that use
the cooperative to collectively conduct their activities (Agricultural co-ops).
Multi-stakeholder – Membership is made up from multiple member categories, including
individuals and/or businesses who join the cooperative to work towards a common aim
(School trusts, leisure trusts, housing associations).
31
Self-Employed – Members are individuals who use the cooperative to collectively support
or conduct their business activities (Freelance cooperatives, collective business services).
Tenants – Members are individuals who rent directly or have shared ownership in a
property or multiple properties (Tenant-run housing cooperatives, residents’ associations).
Workers – Members are individuals who work for and share ownership of the cooperative
(Retail worker cooperatives).
Lack of literature on survival of different cooperative types
Regrettably, there exists very little literature that compares survival rates between different
cooperative ownership classifications. The only types that have been compared are worker
and consumer cooperatives. Sommer et al. (1984) compare the two ownership types.
However, their concern is not associated with survival rates, rather they analyse the
reasons of why persons choose to join either type of co-op. For consumer cooperatives
the main reason seems to be price savings, whereas for worker co-ops the main reason
seems to be the wish for more control over the workplace (Sommer et al., 1984, p. 147).
From this data it is impossible to form hypotheses about the survival length of worker and
consumer cooperatives as well as other types of DF’s and as such this section is focused
more on exploratory analysis rather than the confirmatory analysis seen in previous
sections. As such, no hypotheses are presented.
Methodology and data
To keep hazard ratio data consistent with the previous section of survival analysis, Cox
proportional hazards models are again used to gauge the survival rates of the different
enterprises. The dataset is a modified version of the Co-operatives UK organisations
dataset (Co-operatives UK, 2019a). It is also important to mention that the Co-operatives
32
dataset is limited in the sense that deaths of enterprises had only begun to be recorded
from the year 2011. To adjust for this four distinct samples of co-ops which were created in
years 2011, 2012, 2013 and 2014 will be analysed. This is done, as the dataset only
extends to the year 2019 and to keep analysis consistent, hazard ratios and survival rates
of five years are used to analyse the enterprises. Moving on it is also important to mention
that two of the categories of ownership have been excluded: Co-operatives and Employee
Trust. The former for the fact that it includes umbrella co-operatives that contain others
and as such are not relevant to individual level enterprise analysis. The latter, however,
are excluded due to only containing a sample size of 1, making the category irrelevant to
the analysis. Furthermore, it must be noted that the sample sizes for each of the types of
DF’s is relatively small as the number of co-ops created during each year ranges from 336
to 259. Dividing this small sample into even smaller categories poses the risk of making
the analysis spurious due to small sizes for each of the categories. Table 1 shows the
variables that have been used in this sections’ analysis. Consumer co-ops have been
selected as a reference category in the model since they are a familiar category, that has
been examined in research literature and is easier to interpret.
Empirical model
Table 6 shows the model specifics. It can be seen that in the 2011 cohort there is no
significant relationship between consumer co-op survival rates when compared to different
types, except for worker co-ops which have hazard ratios that are around 4 times higher
than those for consumer firms. This means that a worker co-op started in the year 2011 is
4 times more likely to fail than a consumer co-op on average. The value is highly
statistically significant (p<0.001). When discussing the 2012 cohort the same pattern
emerges with worker co-ops having hazard rations that are around 5 times higher than
those for the reference category on average. The value remains highly statistically
33
significant. There is also now a significant difference between the reference category and
tenant co-ops. The latter is around 3 times more likely to fail than the former on average.
Table 6. Cox proportional hazards model on different co-op survival
2011 2012 2013 2014
Community of 0.779 1.714 1.237 5.067*
Interest (0.294) (0.700) (0.454) (3.767)
Enterprise 1.155 2.241 1.026 5.294
(0.735) (1.517) (0.675) (5.294)
Multi- 1.114 0.959 0.913 3.159
Stakeholder (0.516) (0.466) (0.399) (2.533)
Self- 1.527 2.526 0.726 9.878**
Employed (0.972) (1.710) (0.562) (8.267)
Tenant 0.730 3.126* 1.069 7.941*
(0.552) (1.689) (0.703) (7.941)
Worker 3.908*** 4.812*** 1.499 8.582**
(1.339) (1.932) (0.711) (6.710)
N 319 311 336 259 Exponentiated coefficients; Standard errors in parentheses * p < 0.05, ** p < 0.01, *** p < 0.001
The value is statistically significant (p<0.05). Moving on to the 2013 cohort, it seems that
there is no significant relationship in hazard ratios between consumer and other types of
co-ops. Finally, for the 2014 sample it can be seen that community of interest, self-
employed, tenant and worker co-ops all have higher hazard ratios than consumer firms
(approx. 5, 10, 8 and 8.5 times more likely to fail on average respectively). All values are
statistically significant. After further testing, it seems that all of the models do not seem to
have violated the proportional hazards assumption. Judging from Table 6, it seems that
the only pattern that seems to form is one where worker co-ops are more likely than
34
consumer co-ops to fail. For this, figures for hazard rates and survival graphs are
provided10.
10 The 2013 cohort is excluded due to non-statistically significant results
35
Figure 6. Hazard rates and Kaplan-Meyer survival graph for the 2011 cohort
36
Figure 7. Hazard rates and Kaplan-Meyer survival graph for the 2012 cohort
37
Figure 8. Hazard rates and Kaplan-Meyer survival graph for the 2014 cohort
38
From Figures 6, 7 and 8 it can be seen how each type of cooperative survives and their
hazard rates during each of the years. Consumer firms survive at much higher rates than
worker enterprises. After 5 years in existence, the 2011, 2012 and 2014 cohorts of
consumer DF’s had survival rates of 96%, 91% and 96% respectively. Worker enterprises,
by contrast had survival rates of 65%, 56% and 70% for the same years. As for hazard
rates, it seems that for worker enterprises it increases significantly in the initial 3-4 years of
creation and then goes down. Consumer enterprises, by contrast do not have a steep
increase in hazard rates during the initial five years of creation, however the rate does
increase beyond the five years of creation. The exception being the year 2014, where
hazard rates for consumer cooperatives were only present during the fourth and fifth year
of their life, which suggests that in the early stages, consumer co-ops have almost no
chances of failure. Examining the two co-op types further, it is important to note that the
co-operatives tend to exist in different sectors of the economy.
Table 7. Co-op types by sector (created in years
2011-2014) Sector Consumer firm Worker firm
Agriculture 1 5
Arts and culture 1 7
Digital, Media and communication 3 16
Education 0 12
Energy and Environment 0 2
Finance 9 2
Food service, Accommodation and Pubs 0 7
Health and Social Care 0 9
Housing 0 4
Manufacturing 0 13
Membership associations, Social Clubs and Trade Unions 187 6
Professional and Legal services 0 20
Retail 20 18
Sports and recreation 7 1
Transport 1 1
Other 16 12
Total 245 135
39
Table 7 illustrates this and from the data it can be seen that consumer firms operate
largely as membership associations, social clubs and trade unions, whilst worker firms are
more dispersed in different sectors with the most populous ones being, professional and
legal services, retail, communication and manufacturing. This indicates that there may be
an effect on how survival rates may be affected by sector; however, as this is not part of
the empirical model, it will be addressed in the discussion section.
Discussion: Are industrial sectors responsible for different co-op survival rates?
As there is hardly any literature on different cooperative types, there is not much to go into
in terms of empirical or theoretical results. Examining the sectors in which the two types of
co-ops operate might offer more depth. Table 7 contains the sectors that the 2011-2014
cohorts operated in. What can be seen is that most consumer firms operate as
membership associations, social clubs or trade unions. Worker firms, by contrast, operate
in many different sectors. What is interesting is that consumer firms operate in sectors
which are not as subject to market pressures than worker firms. It seems that consumer
co-ops focus more on connecting persons of similar interests, be it from membership
associations, social clubs or trade unions. Thus, the sectors would not be ones that
operate on a profit-oriented basis, which would not subject them to the randomness of the
market. Worker co-ops, however, operate in sectors where market pressures would be
more severe on a business (i.e. retail, manufacturing, legal services). This shows that
sector may factor into survival rates. As that is a huge topic on its own, it is not feasible to
discuss it here in detail. If market pressures force businesses survive less and consumer
co-ops are not very much affected by it, would it be reasonable to assume that
comparisons between co-ops and capitalist firms would rest more on sector than
ownership? To this the answer is no, as five-year survival rates for worker co-ops, which
40
operate under the same market pressures as regular capitalist firms is still higher (64%11
compared to 43%). This is somewhat offset by high confidence intervals for worker co-ops;
however, the general trend seems to indicate that around a 21% difference between the
two types of firms is observed, which is quite significant in the business world and would
result in around 56 thousand more traditional businesses surviving the five year period in
the United Kingdom if they had the same survival rates as the worker co-ops.
Limitations
As this analysis is exploratory, it was not possible to suggest very complex models that
accounted for different ways co-ops could be affected instead of co-op type. There was
also a small sample size for all cohorts and that has affected confidence intervals quite
drastically, which stifles the accuracy of the model and analysis.
Further research
The preliminary research has proven to be fruitful and as such more research on this topic
is highly encouraged. In the future, adding industry section and other control variables to
achieve more accurate results is recommended. Sample sizes may remain small, so it is
important to note as it may provide a less accurate empirical analysis; however, for the
time being, this seems to be the only way to conduct research into different co-operative
types.
V. Conclusions
This paper has primarily focused on three topics related to co-operatives: their creation,
their comparison to capitalist firms and comparisons between different co-op types.
11 Average for the worker co-op survival rates of the 2011, 2012 and 2014 cohorts
41
The first section tried to view co-op creation as influenced by two factors: unemployment
and disaffection with the current capitalist system. The empirical model found support for
the ideas that both increases in disaffection and unemployment caused a higher growth of
cooperatives. A further discussion was presented that shower that disaffection with
capitalism is caused by events that show that the idea of capitalism being the only viable
system in the world today is challenged (such as the 2008-9 financial crash) and the
inherently exploitative relationship that is created by industrial relations under capitalism,
which can be viewed from either a Marxian or Democratic theory perspective.
Unemployment itself may be a part of the dominant economic system and leads to an
increase of co-operative numbers; however, these two factors seem to be only weakly
collinear.
The second section showed that the 2012 cohort of created businesses and co-ops had
significantly different hazard ratios and survival rates. It was found that on average, co-ops
have a hazard ratio that is 69% lower than that for the traditional business. In terms of
survival, it is seen that 43% of traditional companies survive the 5-year period, whereas
77% of co-ops do. This is a massive difference between the two cohorts and shows that
the co-op model does indeed exist as a more stable alternative to capitalist firms in the
market economy. Reasons for why are bountiful, but the two that have been discussed
were better productivity that is generated by increased co-operation that co-ops offer and
better responses to crises, which prioritise keeping as much of the workforce of the
business employed as opposed to capitalist firms, which can suffer massive layoffs during
crises, affecting workers economically and psychologically.
The third section contributed to the exploratory research around cooperatives and tried to
see how hazard ratios and survival rates differ for different types of co-ops. It was found,
that the only generalisable pattern was between consumer and worker firms. Consumer
42
co-ops performed better survival-wise with their hazard ratios indicating that they are on
average 6 times better at surviving than worker enterprises as well as having a five-year
survival rate of 94% as opposed to worker companies with a 64% five-year survival. The
difference here can be explained by sector, as worker co-ops tend to operate in sectors
where survival is more dominated by gaining profit, whereas consumer co-ops operate
more in sectors that are concerned with organising persons, which means profit may not
be the deciding factor in survival.
All of these topics are promising paths for research into the economy, however, they are
woefully under-research and with this paper it is hoped that a significant contribution has
been made to co-op research as well as that it will inspire more research into the topic.
43
VI. Bibliography
Anonymous, Author, 2020, "Replication Data for: "An empirical analysis of cooperative
creation, survival compared to capitalist firms and survival between different co-op types in
the United Kingdom"", https://doi.org/10.7910/DVN/I35KU4, Harvard Dataverse, V1,
UNF:6:kLQtQtXkAjrnc16oUI/HGA== [fileUNF]
Audretsch, D.B. 1991. New-Firm Survival and the Technological Regime. The Review of
Economics and Statistics. 73(3), p.441.
Barkley Rosser Jr, J. and Rosser, M.V. 2004. Comparative Economics in a Transforming
World Economy. Cambridge, MA: MIT Press.
Ben-Ner, A. 1988. Comparative empirical observations on worker-owned and capitalist
firms. International Journal of Industrial Organization. 6(1), pp.7–31.
Bockman, J. 2011. Markets in the Name of Socialism: The Left-Wing Origins of
Neoliberalism. Redwood City: Stanford University Press
Bruderl, J. and Schussler, R. 1990. Organizational Mortality: The Liabilities of Newness
and Adolescence. Administrative Science Quarterly. 35(3), pp.530–547.
Burdin, G. and Dean, A. 2009. New evidence on wages and employment in worker
cooperatives compared with capitalist firms. Journal of Comparative Economics. 37(4), pp.
517-533
Cheney, G., Cruz, I., Peredo, A.M., Nazareno, E. 2014. Worker cooperatives as an
organizational alternative: Challenges, achievements and promise in business governance
and ownership. Organization. 21(5), pp. 591–603
44
Clark, D. 2019. Unemployment rate of the United Kingdom 2000-2019. [Online]. [Accessed
March 25, 2020]. Available from:
https://www.statista.com/statistics/279898/unemployment-rate-in-the-united-kingdom-uk/
Cleves, M., Gould, W.W., Gutierrez, R.G., Marchenko, Y.V. 2008 An Introduction to
Survival Analysis Using Stata. 2nd edition. College Station: Stata Press
Co-operatives UK. 2019a. Co-operative open data. [Online]. [Accessed 18 November,
2019]. Available from: https://www.uk.coop/open-data
Co-operatives UK. 2019b. Open Data Guide. [Online]. [Accessed 18 November, 2019].
Available from:
https://www.uk.coop/sites/default/files/uploads/attachments/open_data_guide_10.pdf
Cox, D.R. 1972. Regression Models and Life-Tables. Journal of the Royal Statistical
Society. Series B (Methodological). 34(2), pp.187–220.
Craig, B. and Pencavel, J. 1995. Participation and Productivity: A Comparison of Worker
Cooperatives and Conventional Firms in the Plywood Industry. Brookings Papers on
Economic Activity. Microeconomics. 1995(0), pp. 121-174
Dahl. R.A. 1985. A Preface to Economic Democracy. Berkley: University of California
Press
Disney, R., Haskel, J. and Heden, Y. 2003. Entry, Exit and Establishment Survival in UK
Manufacturing. Journal of Industrial Economics. 51(1), pp.91–112.
Doucouliagos, C. 1995. Worker Participation and Productivity in Labor-Managed and
Participatory Capitalist Firms: A Meta-Analysis. Industrial and Labor Relations Review.
49(1), pp.58–77.
45
Eigner, P. and Umlauft, T.S. 2015. The Great Depression(s) of 1929-1933 and 2007-
2009? Parallels, Differences and Policy Lessons. SSRN Electronic Journal
Evans, D.S. 1987. The Relationship Between Firm Growth, Size, and Age: Estimates for
100 Manufacturing Industries. The Journal of Industrial Economics. 35(4), pp.567–581.
Fisher, M. 2009. Capitalist Realism: Is There No Alternative?. Winchester: Zero Books
Fukuyama, F. 1989. The End of History? The National Interest. (16), pp.3–18
Greenberg, E.S. 1986. Workplace democracy: the political effects of participation. New
York: Cornell University Press
International Co-operative Alliance. [no date]. Cooperative identity, values & principles.
[Online]. [Accessed November 20, 2019]. Available from:
https://www.ica.coop/en/cooperatives/cooperative-identity
Kutner, M., Nachtsheim, C., Neter, J. 2004. Applied Linear Regression Models. 4th ed.
New York: McGraw-Hill
Lenin, V. I. 1917. The State and Revolution. [Online]. [Accessed 20 November 2019].
Available from: https://www.marxists.org/archive/lenin/works/1917/staterev/
Marx, K and Engels, F. 1848. The Communist Manifesto [Online]. Moscow: Progress
Publishers. [Accessed 20 November]. Available from:
https://www.marxists.org/archive/marx/works/download/pdf/Manifesto.pdf
Marx, K. 1867. Capital, Volume I [Online]. Moscow: Progress Publishers. [Accessed 30
September]. Available from: https://www.marxists.org/archive/marx/works/1867-c1/
46
Monteiro, N.P. and Stewart, G. 2015. Scale, Scope and Survival: A Comparison of
Cooperative and Capitalist Modes of Production. Review of Industrial Organization. 47(1),
pp.91–118.
Muller, B. 2019. What is Austerity and How Has It Affected British Society?. The New York
Times. [Online]. 24 February. [Accessed 20 November 2019]. Available from:
https://www.nytimes.com/2019/02/24/world/europe/britain-austerity-may-budget.html
Mustafa, A. 2011. London faces a summer of strike chaos over austerity. Gulf News.
[Online]. 29 June, [Accessed 20 March 2020]. Available from:
https://gulfnews.com/business/analysis/london-faces-a-summer-of-strike-chaos-over-
austerity-1.829578
Office for National Statistics. 2018. The 2008 recession 10 years on. [Online]. [Accessed
20 March 2020]. Available from:
https://www.ons.gov.uk/economy/grossdomesticproductgdp/articles/the2008recession10ye
arson/2018-04-30
Office for National Statistics. 2019. Business Demography – 2017. [Online]. [Accessed
November 18, 2019]. Available from:
https://www.ons.gov.uk/businessindustryandtrade/business/activitysizeandlocation/dataset
s/businessdemographyreferencetable
Perotin, V. 2004. Early cooperative survival: the liability of adolescence. Advances in the
Economic Analysis of Participatory & Labor-Managed Firms. 8, pp.67–86.
Sheather, S. 2009. A modern approach to regression with R. New York: Springer
Sommer, R., Schlanger, D., Hackman, R. and Smith, S. 1984. Consumer Cooperatives
and Worker Collectives: A Comparison. Sociological Perspectives. 27(2), pp.139–157
47
Stensrud, M.J. and Hernán, M.A. 2020. Why Test for Proportional Hazards? JAMA.
Tajfel, H. and Turner, J. 1979. An Integrative Theory of Intergroup Conflict. In: Austin, W.
G. and Worchel, S. eds. The social psychology of intergroup relations. Monterey:
Brooks/Cole
Zirakzabeh, C.E. 1990. Theorizing about Workplace Democracy Robert Dahl and the
Cooperatives of Mondragón. Journal of Theoretical Politics. 2(1), pp.109–126.
48
VII. Appendix
Table 1. Variables used during the analysis and their descriptions
Section Variable Label Type Description
III coopscreated Number of Co-ops created Dependent The number of co-operatives created during a given year
III unemployment Unemployment increase Independent Unemployment increase during a given year
III aftercrisis Disaffection with capitalism
Independent
Years before, during or after the 2008-2009 economic crisis. Higher value denotes higher disaffection with the current economic system
III year Year Contextual Denotes the given year for the data
IV diedonyear Died on year Dependent Denotes the year of the lifespan during which the enterprise died
IV regstat Registration status Dependent Denotes whether the enterprise is alive (1) or dead (0)
IV organisation Cooperative Independent
Denotes the type of organisation an enterprise is. 0 for "Capitalist firm", 1 for "Cooperative"
V diedonyear Died on year Dependent Denotes the year of the lifespan during which the enterprise died
V regstat Registration status Dependent Denotes whether the enterprise is alive (1) or dead (0)
V ownership Ownership Classification Independent Denotes the ownership classification of a cooperative
V yrcreated Year created Contextual Denotes which year the enterprise was created on
VIII. Notes
Data and the do files for the replication can be found via this direct link:
https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/I35KU4